Research On The Soft Tissue Deformation Modeling Methods In Virtual Surgery

With the development of computer science and the improvement of clinical medicine,virtual surgery system has become an essential assistant tool in many aspects,such as surgical training,intraoperative navigation,surgery scheme planning and surgical outcome prediction.However,the research of virtual surgery system belongs to an interdisciplinary frontier technology,which makes the corresponding study extremely difficult.As the core part of a virtual surgery system,the realistic soft tissue deformation modeling method plays a significant role in overcoming the restrictions of traditional surgical training method,developing optimal surgical scheme as well as improving the surgery success rate.Therefore,this paper mainly studies the soft tissue deformation modeling method in virtual surgery system,and makes an in-depth study of the visual authenticity and the real-time performance of soft tissue deformation.The main work is as follows:Firstly,this paper studies the biomechanical properties of human soft tissue and describes several widely-used viscoelastic models,which laid the theoretical foundation for further study of realistic soft tissue deformation modeling.In the aspect of soft tissue geometric modeling,this paper proposes a bounding box based texture coordinates calculation algorithm.It can be treated as an efficient way to address the problem of poor visual authenticity of human soft tissue model obtained by Amira software.Model data storage manner is also studied to make the reconstructed soft tissue model reusable.In addition,this paper also studies the Sphere and AABB based hybrid hierarchy bounding box tree to accelerate the intersection detection between virtual surgical instrument and soft tissue.Corresponding experimental results show that the proposed hybrid hierarchy bounding box method can improve the efficiency of collision detection.Finally,this paper introduces haptic feedback technology in virtual surgery system as well as two feedback force calculation methods.Compared with the spring-damper model based force calculation method,texture based force calculation method can obtain more realistic haptic feedback effect.Secondly,this paper studies mass-spring model based soft tissue deformation method,and introduces self-springs in mass-spring system to overcome the problem of collapse and super elastic when only concerns structure and bending springs and to improve the deformation authenticity.Additionally,this paper also studies several popular numerical integration methods during soft tissue deformation calculation.Mass-spring based experimental results show that implicit Euler integration method remains relative high numerical accuracy even with a larger time-step.In order to overcome the problem of poor real-time deformation of global mass-spring model,this paper proposes a collision detection technique based local adaptive mass-spring model,and studied the reversed bounding box update algorithm of the hybrid hierarchy bounding box tree in the local deformation area.Stomach model based experimental results show that local adaptive mass-spring model can reduce the computational overhead greatly,and further enhances the real-time performance of the system without losing deformation authenticity.Thirdly,this paper conducts an in-depth study of co-rotational finite element method(FEM)and introduced two tetrahedral tools which can generate volumetric model from surface model.Since the calculation complexity of co-rotational FEM based deformation method is relative high,this paper studies multi-resolution technology based deformation acceleration method to make the real-time deformation simulation of soft tissue with high complexity possible.In addition,the paper also studies the extraction method of tetrahedral rotation matrix during the deformation calculation process.Adaptive rotational matrix extraction according to the deformation degree is proposed to overcome the problem that polar decomposition can hardly achieve real-time because of its huge computation overhead.Polar decomposition method is used to guarantee the deformation authenticity of the tetrahedron which has large deformation,while orthogonal decomposition method is used for those with small or without deformation to guarantee the real-time performance of the system.Heart model based experimental results show that the proposed adaptive rotational matrix extraction method according to the deformation degree of soft tissue deformation can complete the tetrahedral rotation matrix extraction process rapidly with high accuracy deformation.Finally,the vascular deformation in master-slave based vascular interventional surgery training system is studied.Since the vascular model owns too many tetrahedrons,its computation overhead slows down the efficiency of collision detection.Therefore,this paper proposes a hybrid vascular geometric model by combining surface model and volume model together,which processes collision detection and deformation calculation separately.Furthermore,this paper also studies position based vascular deformation method,and deduces the volume conservation constraint of tetrahedron to solve the penetration problem of vascular model during the deformation simulation.Since introducing volume conservation constraint in position based deformation calculation increases the computation overhead greatly,this paper proposes a spatial partition based spatial accelerating method to accelerate vascular deformation calculation to guarantee the real-time performance of numerical solving.The simulation results show that conservation constraint can solve the penetrating problem of vessel model during deformation.Meanwhile,the spatial partition based rapid deformation method can improve the real-time performance without losing deformation authenticity.